Assessment of Imja Glacier Lake Outburst Flood (GLOF) Risk in Dudh Koshi River Basin Using Remote Sensing Techniques Kamal P

Home , Dudh Koshi, Imja Glacier, Imja Khola, Imja Tsho, Lhotse Shar Glacier

December 2010 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel 75 Assessment of Imja Glacier Lake outburst Flood (GLOF) Risk in Dudh Koshi River Basin using Remote Sensing Techniques Kamal P. Budhathoki1, O. R. Bajracharya1 & B.K.Pokharel1 1 Department of Hydrology and Meteorology, Babarmahal, GPO Box 406, Kathmandu, Nepal

Abstract

Glacier lakes are common phenomena in the Himalaya region of Nepal. Glacier lake outburst floods have repeatedly caused the death tolls and severe damage to downstream infrastructures. In Himalayas, a vital uncertainty about the glacier lake hazard potential still exists, thereby the effects of accelerating rates of glacier retreat and expansion of Glacier Lake could be the wake of recent global warming and resulting climatic changes. The paper, first describes the general different-level approach upon which the study is based. Then, in the methodological part, applicable remote sensing techniques, geographic information system (GIS) and statistical methods are presented. Observed data of lake area, volume, and depth having similar lake characteristics reported in the different literature are used to develop empirical equations by using statistical methods. The values of r2 (coefficient of determination) - obtained are very high (r2=0.939 for depth – area relationship and r2= 0.990 for volume – area relationship). The comparison of the empirical expression clearly indicated that there is more than 90% variation in the dependent variable, lake volume,as explained by the linear regressions in both cases. Area of Imja glacier lake for different years are determined from the available satellite imagery and volume of the Imja glacier lake are estimated using the expression: V = 0.094A1.453.developed from linear regression analysis of the observed data. Similarly, mean depth can be estimated by using the expression: D = 0.94A0.452.

After the preparation of maps and data, a scheme of decision criteria for the evaluation of hazard potential of Imja glacier lake is established. A list of decision criteria is a documented set of factors that are used to examine and compare for evaluating the hazard potential of a glacier lake. The empirical scores are given in terms of hazard magnitude for hazard rating. Analysis of Imja glacier lake based on the empirical scoring system clearly indicated that GLOF risk of the possible outburst from Imja glacier lake is MODERATE. A systematic application of remote sensing based methods for glacier lake outburst flood risk assessment is applicable and thus recommended.

Keywords: Glacier lake outburst, remote sensing, risk assessment, hazard potential, empirical parameters, climate change

1. INTRODUCTION The studies also reflected that Himalayan glaciers have undergone various stages of retreat, advance Climate change studies showed that the Himalayan and standstill in those years, while the rate of glaciers have been receding since last 150 years. glacier retreat has been more rapid since 1990’s.

Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal 76 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel December 2010 Results of such phenomena have been responsible them are found to be a considerable size. Data used for the development and expansion of glacier lakes in the study were high and low resolution satellite in the Himalayas. Most of the glacier lakes in Nepal data and past study reports. The paper first describes Himalaya are likely to have begun to develop in the the general different-level approach upon which the 1950s (Watanabe et al., 1994; Sakai, 1995). Many of study is based. Then, in the methodological part, the glacier lakes are formed on the debris-covered applicable remote sensing techniques, geographic areas of glaciers. Some of the glaciers have higher information system (GIS) and statistical methods rates of retreat due to which the glacier lakes are are presented. After the preparation of maps and increasing at unexpected rates. The process of fast data, a scheme of decision criteria (ICIMOD, 2001; glacier retreat and expansion of glacier lakes could Huggel et al. 2002 and RGSL, NEA. 2004) for the be the wake of recent global warming and leading evaluation of hazard potential of Imja glacier lake climatic changes. is discussed. It is crucial and challenges to judge and evaluate the potential risk assessment without Glacier lakes that are dammed with unconsolidated having the detailed geophysical field investigation weak natural moraines pose potential risk of their works. catastrophic failure, causing a huge flash flood in the downstream areas commonly known as Glacier 1.1 Socioeconomic impacts Lake Outburst Flood (GLOF). GLOF is a relatively of glacier lake outburst new type of natural hazard in the Himalayan floods region which has emerged as consequence of ongoing climate change. Glacier lakes drew the GLOF events are severe geo-morphological attention of many researchers, scientists and hazards and their floodwaters can wreak havoc on concerned institutions and abroad following the all human structures located on their path. Much outburst of Dig Tsho Lake in eastern Nepal on 4 of the damage created during GLOF events is August 1985. In the past couple of decades, several associated with the large amounts of debris that lakes caused outburst floods affecting human lives accompany the floodwaters. Damage to settlements and infrastructure (Vuichard and Zimmermann and farmland can take place at very great distances 1987; Yongjian and Jingshi 1992; Hanisch et al. from the outburst source (WECS 1987b). 1996). Past field investigation have identified that few glacier lakes are potentially dangerous and Direct and indirect sociological impacts are the recommended to adopt appropriate mitigation loss of human lives and the agricultural lands measures to minimize the possible damage in the when converted to debris filled lands due to which event of a GLOF. Most glacier lakes have not yet villages have to be shifted to other safe locations. been identified or studied because of their remote Based on the past GLOF events, many reports locations. coated that once every three to ten years, a GLOF has occurred in Nepal with varying degrees of The study especially focused on the use of remote socioeconomic impact. Brief studies of GLOFs sensing technology and other methods applied in throughout the world showed that there are no Imja glacier lake for potential risk assessment. Imja simple direct means of estimating the recurrence lake is one of the head sources of Dudh Koshi river. of GLOFs. Various GLOF events have occurred There are many glacier lakes in Dudh Koshi river periodically over the past in Nepal. Most of them basin. Imja glacier lake is a bigger one and few of were not recorded. An early overview of glacier

Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal December 2010 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel 77 lake outburst hazards in Nepal was given by Ives A sudden burst of a glacier lake in any sub-basin (1986). Few of them were documented based may cause loss of lives and damage to the villages, on information from inhabitants living in the cultivated lands and infrastructures (hydropower, Himalayan highland along the river courses. Many irrigation sites etc.) situated downstream. Debris of these events originated in Tibet autonomous deposited by flood , landslide and temporary region of China outside the political boundary of damming due to flood can cause the major to the Nepal. people living hindrances downstream.

As a consequence of the GLOF events, attention A sudden burst of Imja lake will cause extensive was drawn to large glacier lakes of Nepal and damage along the entire length of the Imja Khola/ Tibet. Few rivers of Nepal have their origin at Dudh Koshi valleys down to the confluence with Tibet, such as Arun river, Tama koshi, Sun Koshi, Sun Koshi, an overall distance of about 90 km. Gandaki and Karnali etc. Highly awareness Study showed that (Shrestha, 2007) a minimum and the realization of the destructive nature of estimated outburst flood from Imja Glacier Lake GLOF, the necessities of detail investigation of would affect significant changes on the local potentially dangerous lakes would come into topography and the river courses. In addition, the higher priority. International practices had shown low lying terraces of Dingboche village, Shanjo that if preventive measures are taken in time in Kalka and the hotels and lodges along trekking order to avoid likelihood disastrous flood that routes to Chhukhung, would be washed away. could occur from Glacier Lake, lots of lives and Tsuro Og, Orsho, Pangboche, Phunki, Thumbug, properties could be saved. The first mitigation Nyambua, Phakdingma, etc. are the villages program against GLOF in Nepal was undertaken situated within 32 km downstream of the lake in Tsho Rolpa, the largest glacier lake in Nepal. and these villages are most vulnerable to possible Selection of a mitigation measure against GLOF GLOF from Imja Glacier Lake (Braun and requires in depth knowledge in different aspects Fiener, 1995). The study was carried out by using of the glacier lakes. hydrodynamic model coupled with extensive use of geo-informatics. The model outputs provide 1.2 Importance of GLOF Risk information on flood arrival time, discharge and assessment in Dudh Koshi depth at different locations in the valley which river basin could improve the understanding on the effects of a GLOF from Imja Lake. The study of glacier lakes is very imperative for the planning and implementation of any 2. pHYSICAL setting of Imja socioeconomic and water resource development glacier lake in Dudh Koshi project. Past records reveal that glacier lakes have basin created devastating floods and damage to major constructions and infrastructure. Imja Lake is located in Solukhumbu district, Sagarmatha Zone, Eastern Development Region There are a lot of hydropower potential sites as of Nepal at aerial distances of about 163 km well as thousands hectors of cultivated lands along from Kathmandu and about 6 km south of Mount the Dudh Koshi river courses. Dudh Koshi basin Everest (Figure 1). The lake is an intra-glacier has of many glaciated sub- basin. lake in the Imja glacier and its two major tributary

Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal 78 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel December 2010 glaciers are Lhotse Shar glacier flowing from Lake are very steep and the loose surface materials northeast and Ambulapcha glacier from south. continuously falling into the lake surface. The end It is located south east of Chokarma Tsho and moraine is relatively lower than the later moraines. northeast of Ambulapcha Tsho. The valley opens The end moraine rises sharply to the height of in the southwest. The latitude and longitude of the approximately 100 m from the riverbed level of lake are 270 59’ 17’N and 860 55’ 31”E. The lake Imja Khola (Yamada, 1998). is situated at an elevation of approximately 5010 m (a.m.s.l). The major drainage system of upper Imja khola basin is shown in figure 2. There is only one outlet flowing through the southern part of the end moraine dam. The melt water drains into the lake from en-glacier channels (WECS, 1991). The lake collects water from drainage basin area of 37.67 km2 of which 69 % is glacier covered. The glacier lake is strongly influenced by the sub- tropical monsoon climate (June-September), which produces a pronounced summer precipitation, greater than 80% of annual total.

The moraine dam is composed of unconsolidated Figure 1: clay, silt, sand, gravel and boulder, making the dam Location map of the Dudh Koshi basin very unstable. The lateral moraine surfaces of Imja

Figure 2: The river system in upper Imja Khola watershed

Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal December 2010 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel 79 3. Materials and accurate results for the lake area measurement. Methodology Different dates are used to monitor the changes. The generated DEM combined with satellite The basic materials required for this study are images helped in deciding for discrimination of satellite images, aerial photographs and past features and land-cover types for better perspective study reports. All the components of cryospheric viewing. DEM itself is used to create various data region cannot be measured directly from space sets on the lake and glaciers (e.g., slope, aspect). but some parameters such as glacier lake, glacier DEM should be compatible with and of reliable area, terminus position, transient snowlines and quality when compared with other data sets. The surface elevations can be extracted from airborne satellite images draped over the DEM for better and space-borne scanning. Medium resolution interpretation work. satellite data (10 - 90 m) have become available for cryospheric studies since the early 1970s, with For lake detection, the following techniques are the launch of new spaceborne sensors: Landsat based on optical panchromatic and multispectral Multispectral Scanner (MSS), Landsat Thematic remotely sensed data. The detection of glacier Mapper (TM) Enhanced Thematic Mapper Plus lake area using multispectral imagery involves (ETM+) and the Advanced Visible and Near discriminating between water and other surface Infrared Radiometer type 2 (ALOS: AVNIR-2). types. Delineating surface water can be achieved Advanced Land Observing Satellite (ALOS) using the spectral reflectance differences. Water launched on January 2006 by the Japanese Earth strongly absorbs in the near- and middle-infrared observing satellite data such as the Advanced wavelengths. Vegetation and soil, in contrast, Visible and Near Infrared Radiometer type 2 have higher reflectance in the near and middle- (AVNIR-2). Images of course resolution (Landsat infrared wave lengths, hence water bodies appear MSS, TM and ETM+ of 1970s, 80s, 90s and 2000) dark compared to their surroundings when using and image of AVNIR-2 (10 m) of 2008 were used these wavelengths (Pietroniro and Leconte 2000). for mapping, monitoring and assessing the GLOF Applying the idea of two spectral channels with risk of Imja glacier lake. The activity of this lake maximum reflectance difference for an object in terms of hazard score is developed by using the (i.e., water), a blue channel (maximum reflectance logical calculation in the GIS, statistical analysis, of water) and a near-infrared (NIR) channel decision criteria and empirical scoring. (minimum reflectance of water) were chosen.

3.1 Satellite images and Image Different image enhancement techniques were used classification to identify the lake and the glaciers contact with them. In this regard visual interpretation methods Remote-sensing data like MSS, TM, ETM+, with the knowledge of image interpretation keys: ALOS (AVNIR) for different dates are used to color, tone, texture, pattern and association, prepare the maps of Imja glacier lake for GLOF shape, shadow, etc and experience of the terrain risk assessment. Due to higher spatial resolutions conditions played a vital role for glacier lake risk and freely available, TM and ETM image are used assessment tasks. Analyzing by different spectral as the data source with least cloud cover. The band combinations in false color composite (FCC) combination of digital satellite data and the Digital and in individual spectral bands helped to identify Elevation Model (DEM) is used for better and more the surrounding situation of the glacier lake.

Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal 80 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel December 2010 Different color composite images were used to Multi-temporal data sets are an invaluable source simplify the different land-cover features. Water for the comparison with past, in between and present bodies were first separated from snow, ice and situations. For instance, a number of satellite data vegetation by applying several types of supervise before and recent enabled mapping and better classification methods. Misclassified areas of understanding of mother glaciers, glacier lake and shadow were identified and interpolate the values its changes with degrading nature. Thus, multi- based on past data and on field knowledge of temporal applications are of particular important topography of glacierized area. Frozen water for monitoring and assessing glacier lake outburst detected by the automatic classification as snow, flood risk. Accurate co-registration of the repeat so these areas were also added manually. Glaciers, data, i.e. identical ground co-ordinates for identical snow cover hampered the demarcation of lake (stable) terrain points, is a crucial prerequisite of area, especially in the shadowing region of the any image change detection within a certain span basin. of time (Figure 3).

Landsat MSS, 1976 Landsat MSS, 17 Dec.1991

Landsat TM, 30 Oct.2000 ALOS AVNIR-2, 24 Oct. 2008

Figure 3: Imagery of the Imja Glacier Lake area of different dates

Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal December 2010 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel 81 4. Mapping, monitoring and were applied. The glacier lake on each image was computing volume of Imja delineated by the help of DEM. The 3D map of glacier lake Imja glacier lake based on satellite image of 2008 is as shown in figure 4. The longitudinal growth of the lakes is not only due to the melting of the cliff-shaped terminus but with the collapsing of the cliff (calving process) together the degradation processes in and on the surface of end and lateral moraines. For the delineation of the exact contact points between the lakes and the terminus of the mother glacier, the high resolution imageries are used. The study was mainly emphasized on the time series mapping of glacier lake and monitoring of the surface area changing. Shifting backward of the upstream end of the lake and changing on the end moraine morphology at the surface of ice cored-debris Figure 4: Downstream view of 3D image of upper covered end moraine in front are remarkable Imja river basin (24 Oct, 2008) causes to expand the lake. 4.2 Empirical methods for lake 4.1 Mapping and monitoring of volume estimation glacier lakes Surface area of the lake could be extracted from The mapping and monitoring of glacier lake remotely sensed data. However, glacier lake are totally based on different types and dates of volume rather than area is essential to estimate satellite images, secondary data source and aerial potential peak discharge for a possible outburst. photograph. The data set images of 1980s, 1990s, There is no realistic way to directly derive the 2000 and 2008 of Dudh Koshi basin were used. volume from optical remote sensing data despite Least snow cover and cloud free satellite images a number of efforts to perform mapping of surface were selected for lake area measurement. Least water and depth measurements from satellite snow cover in the Himalayas occurs generally in image (Benny and Dawson 1983). The reliability the summer season (May-September). But during of these remote sensing based bathymetric studies this season, monsoon clouds will block the views. depends on finding a significant relationship If snow precipitation is late in the year, winter between water depth and reflected energy images are also suitable except for the problem of (Baban 1993). Large number of glacier lakes in long relief shadows in the high mountain regions. poorly known high mountain areas makes this The frozen lake always has a level at the toe of approach unfeasible to calculate the volume, so an the Imja glacier tongue. Mapping of the lake was empirical approach is chosen instead by Huggel carried out visually as well as digitally by the et al. (2002) for the Swiss Alps. They used the help of DEM. In both the visual interpretation empirical models for outburst characteristics by and digital feature extraction techniques, the computing lake volume and outburst discharge. analyst’s experience and adequate field knowledge Similar attempt has been made to develop the

Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal 82 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel December 2010 empirical relation for computing the volume TAR and Nepal Himalayas), and the functional of glacier lakes. Hence, observed data of lake empirical relationship are developed for volume area, volume, and depth reported in the different - area and area – depth by applying statistical literature (Table 1) are compiled (Swiss Alps, regression analysis.

Table 1: Compiled data on glacier lake area, volume and mean depth

S.No Name of glacier Area (m2) Volume (m3) Mean depth Type of lake Reference lake (m)

1 Gruben Lake 5 10000 50000 5 Thermokarst Teysseire 1999 2 Gruben Lake 1 23000 240000 10.4 Moraine-dammed KÍääb 1996 3 Lac d' Arsine 59000 800000 13.6 Moraine-dammed Vallon 1989 4 Nostetuko lake 262200 7500000 28.6 Moraine-dammed Clague &Evans 1994 5 Gjanupsvatn 600000 20000000 33.3 Ice-dammed Costa & Schuster 1988 6 Abmachimai Cho 565000 19400000 34.3 Moraine-dammed Meon &Schwarz 1993 7 Gokyo Lake 400000 21000000 40 Ice-dammed WECS 1986 8 Gelhaipu Cho 580000 25500000 42 Moraine-dammed Xu & Feng 1994 9 Thulagi 760000 31800000 42 Moraine-dammed WECS & JICA 1996 10 Lower Barun 600000 28000000 47 Moraine-dammed WECS 1993 11 Imja Tsho 750000 33480000 48 Moraine-dammed DHM 1999 12 Tsho Rolpa 1390000 76600000 55.1 Moraine-dammed WECS 1994 13 Quangzongk Cho 760000 21700000 29 Moraine-dammed Xu & Feng 1994

100

10 Mean Depth, m

1 1000 10000 100000 1000000 10000000 Surface Area, m2

Figure 5: Regression between mean depth and area for ice - and moraine – dammed lakes

The regression analysis between area (A) and mean depth (D), yields the following equation

D=0.094A0.452 ------(Eqn.1), where r2 = 0.939

Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal December 2010 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel 83

10000000 10000000 3 1000000

Volume m Volume 100000 10000 1000 1000 10000 100000 1000000 10000000 Surface Area, m2 Figure 6: Regression between volume and area for ice - and moraine – dammed lakes

The glacier lake volume (V) can be estimated using following regression equation.

V = 0.094A1.453------(Eqn.2), where r2 = 0.990

A similar expression was defined by the Canadian Inland Water Directorate for glacier-dammed lakes (Evans 1986a):

V = 0.035A1.5 ------(Eqn.3),

4.3 Surface area and volume relationship). The comparison of the empirical computation expression clearly indicated that there is more than 90% variation in the dependent variable, (lake Imja Glacier Lake started to grow with size rapidly volume,) as explained by the linear regressions. around year 1963. Surface area of Imja Glacier Lake was found increasing up to the date 24 The results of an analysis of variance (ANOVA) October 2008 (Table 2). Using regression equation showed a large regression sum of squares (5.650 2, volume of Imja glacier lake has been estimated. for depth-area and 58.310 for volume- area) in Glacier Lake was found increasing up to the comparison to the residual sum of squares (0.365 date 24 October 2008 (Table 3). Using regression depth-area and 0.564 for volume-area ) indicating equation 2, volume of Imja glacier lake has been the model accounts for most of variation in the estimated. dependent variable. Very high residual sum of squares indicate that the model fails to explain 4.4 Interpretation of the a lot of the variation in the dependent variable, result and hence necessary to look for additional factors that help account for a higher proportion The values of r2 (coefficient of determination) of the variation in the dependent variable. The obtained are very high (r2=0.939 for depth – area significance value of the F statistic is zero (smaller relationship and r2 = 0.990 for volume – area than 0.05) then the independent variables do a Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal 84 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel December 2010 good job explaining the variation in the dependent volume from satellite derived area of glacier lakes. variable. This method uses the area (A) of a lake, which is easily measured accurately in the satellite imagery Water volume contain in the lake is one of the most and to convert to a volume (V) using the expression: important factors for the potential hazard criteria V = 0.094A1.453. The estimated volumes (Table 3) of GLOF. Very little information and unavailability of Imja glacier lake at different dates are thought of required data of the study area, the empirical to be accurate to the level of ±15% after validation relation played a significant role for computing the with few observed data.

Table 2: Surface area and computed volume of Imja Glacier Lake

Year Area, km2 Area, m2 Volume, m3 Source 1963 0.03 30000 297788 Prof. Fritz Muller, member of Swiss Everest/Lhotse expedition, 1956; topo map (1978) & surveyed (1956-1963) 1975 0.3 300000 8431560 Aerial oblique photograph, Japanese Glaciological Expedition of Nepal (GEN),1975 1978 0.36 360000 10986999 Aerial oblique photograph, Japanese Glaciological Expedition of Nepal (GEN), 1978 1984 0.47 470000 16181320 Vertical photograph (National Geographic Magazine, 1984), Landsat image, 29 Oct. 1984 1986 0.53 530000 19265328 SPOT 1, HRV 1 Image, 23rd March 1986 1991 0.652 652000 26026476 Landsat TM, Dec 1991( DHM and Central Dept of Hydrology and Me- teorology (CDHM), TU, 1998) 1992 0.685 685000 27960862 Aerial Photograph, Topographical Survey Branch, Survey Department, HMG/Nepal, WECS, 1992 1996 0.721 721000 30119646 Topographical map, 1996, and field survey (DHM & CDHM, TU) 1997 0.74 740000 31278960 DHM/Nepal & Hokkaido University/Japan, March 1999 1999 0.75 750000 31894574 DHM, 1999 2001 0.788 788000 34267618 Dr. Tomomi Yamada, ASTER AVN color composite 2004 0.889 889000 40825660 ASTER AVN, 21Nov.2004 (DHM) 2008 0.9595 959500 45609694 ALOS AVNIR-2, 24 Oct. 2008 (DHM)

5. Decision criteria for retreat of the glacier affect the hydrological identification of potential regime between the present day glacier and danger the lake dammed by the moraines. Sudden natural phenomena with a direct effect on a The glacier lakes located away from present- lake, like ice avalanches or rock and lateral day glaciers and the downstream banks are moraine material collapsing on a lake, cause usually made of bedrock or covered with a moraine breaches with subsequent lake outburst thinner layer of loose sediment generally do events. Such phenomena have been well known not pose an outburst danger. On the other in the past in several cases of moraine-dammed hand, the moraine-dammed glacier lakes have lakes, although the mechanisms that play are the potential for bursting. The advance and not fully understood.

Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal December 2010 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel 85 Field investigation and inventory in middle 8. Moraine dam condition (ice-cored/ section of the Tibetan Himalayas showed that the thermokarst) dangerous glacier lakes exhibit the well closed 9. Supra- and en-glacier channel basins with high and narrow moraine ridges and 10. Complex/compound threat (height of contact with the mother glaciers which could glacier snout, glacier with deep crevasses, really be marked a potential outburst danger. slopes, orientation, dipping, etc) The major triggers of the outburst are (i) ice avalanche from advanced glacier tongue with a Based on the above hazard/ scoring indicators/ volume usually at millions of cubic meters and criteria, following sets of decision criteria (Table 3) (ii) ablation activities of dead ice beneath moraine were prepared and applied to Imja glacier lake, ridges (Xu and Feng, 1994). because the lake has been subjected to closer examination. Two different scale levels rather It is huge and very difficult task to establish than three (Huggel et al, 2002) exemplify the basic a set of criteria without doing more detailed approach of this study: investigation on geo-technical, glaciological and geo-morphological as well as hydrological aspects Level 1: The first level comprises the basic in and around the potential dangerous glacier lake detection of Imja glacier lake in Imja basin and area. Unfortunately, there is no common agreed its high spatial and temporal variability from a set of indicators to assess the risk of GLOFs from powerful tool like space-borne remote sensing young, moraine dammed glacier lakes (Grabs and techniques. Hanisch, 1993). However, the criteria developed by ICIMOD for identifying the potentially dangerous Level 2: This level assesses the hazard potential glacier lakes are prerequisite for hazard/scoring of the lake detected in Level 1. Level 1 was system for moraine-dammed lake outburst complemented by information about the related potential. The decision criteria for evaluating a hazards. Therefore, in Level 2, image interpretation, glacier lake’s hazard potential in the Himalayas analysis and GIS modeling based on multisource are mainly outlined briefly as: data such as high resolution satellite imagery (AVNIR: 10 m) and digital elevation models were 1. Surface area and volume of water stored; applied. 2. Lake in contact with active ice body of a glacier (other than parent glacier); Level 3: This level is mainly concerned with 3. Steeper slope of the moraine walls detailed investigations of lake recognized through and mass movement or potential mass Levels 1 and 2 to have a significant hazard movement in the inner slope and/or outer potential. Typically, such investigations are slope concerned with one specific lake and apply very 4. Risk of ice calving from glacier snout/ice high resolution remote sensing data, geophysical cliff studies, and other field work. Application of 5. Freeboard relative to the lake water level Level 3 is usually a prerequisite for carrying out 6. Rock fall/ice avalanche in the periphery of on-site mitigation measures and is not applicable the lake in the present risk assessment of glacier lake 7. Evidence of strong dam seepage outburst.

Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal 86 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel December 2010

Table 3: Decesion criteria for identifying potentially dangerous glacier lake

Decision criteria Scale level Appropriate techniques and data

Surface area and Volume of water/ Growth rate 1, 2 Computed from time series satellite images and derived volume from empirical relation/ change detection Seepage through the moraine dam 2 Change detection: satellite image time series Supra- and en–glacier channel 1, 2 Visual recognition from high resolution satellite image and supervise image classification Risk of ice calving from glacier snout/ice cliff 2 Using contour and slope from generated DEM Rock fall/ice avalanche in the periphery of the 2 Visual recognition from satellite images ( (10 m of AVNIR) lake with the help of generated slope and contour maps. Freeboard relative to the lake water level 2 Stereo Satellite image and photogrammetric techniques including dam height (contour and slopes), but field visit is essence. Moraine dam condition (ice-cored/thermokarst, 2 Visual recognigation from satellite image of 10 m resolution seepage, steeper slopes etc) of AVNIR with the help of generated slope and DEM maps. Complex/compound threats (height of glacier 1,2 AVNIR satellite data classification, DEM and contour snout, glacier with deep crevasses, slopes, analysis and visualization techniques, but field visit is orientation, dipping, etc ) essence.

6. gloF risk assessment/ volume, dam condition, glacier condition and the analysis surroundings of the lake environment such as hanging glaciers, rock fall and snow avalanches A methodology to determine the nature and etc (triggering mechanism) in a reasonable time extent of risk by analyzing potential hazards and interval. evaluating existing conditions of vulnerability that could pose a potential threat or harm to infra- 6.1 eMPirical scoring system structure, people, property, livelihoods and the for GLOF hazard potential environment on which they depend. The process of conducting a risk assessment is based on a Hazard score based on criteria was developed review of both the technical features of hazards for hazard assessment of glacier lakes by RGSL such as their location, intensity, frequency and (1998) during a GLOF study in Bhutan, and tested probability; and also the analysis of the physical, in Nepal, Peru and Bhutan. RGSL also developed social, economic and environmental dimensions of an empirical scoring system (GLOF Source vulnerability and exposure, while taking particular Parameter) for moraine-dammed lake outburst account of the coping capabilities pertinent to the hazard for GLOF Risk assessment of Rongxer risk scenarios. basin (RGSL, NEA 2004). The scheme ranks a series of empirically derived criteria that affect Hence, assessment of GLOF risk is a very difficult magnitude and probability of an outburst. Mostly, and complicated work. Important criteria is the similar approach is applied in the present study. rate and nature of change and these can only be Criteria affecting hazard/ score are established ascertained through subsequent monitoring and based of the decision criteria, scale level and re-assessment of the surface area and lake water appropriate techniques & data for evaluating a

Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal December 2010 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel 87 glacier lake’s hazard potential in the Imja basin of m3 gave a score of 2 for parameter 1 ( ID 1); if the Nepal Himalayas. Empirical scoring system is more range around 5x106 to 1.6 x107 m3, a score of 10; practical and thus helped to rank the magnitude of and if much larger 1.6 x107 m3 then a score of 40 high-risk to lower orders in terms of glacier lake’s is given. The same system is applied by different hazard potential for moraine-dammed lake. The way for each parameter of different ID and the numbers are given in terms of hazard magnitude score is added together to give a single value. for hazard rating. The method ranks a series of This value is then compared with an empirical empirically derived decision criteria that affect scale indicating potential hazard (Table 5). Higher magnitude and probability of an outburst (Table 4). scores indicate potential of greater hazard and prioritize accordingly. If the total score exceeds Here the score is developed, as for example, a lake 90, then an outburst is considered to be likely at water volume whose volume is less than 5x106 any time.

Table 4: Empirical scoring system for Imja GLOF hazard

ID Criteria affecting hazard/ score 0 2 10 40 1 Surface area and Volume of water N/A Low Moderate Large 2 Seepage through the moraine dam None Minimum Moderate Large 3 Supra- and en-glacier channel None Low Moderate Large 4 Risk of ice calving from glacier snout/ice cliff N/A Low Moderate Large 5 Rock fall/ice avalanche in the periphery of the lake N/A Low Moderate Large 6 Freeboard relative to the lake water level including dam height N/A Low Moderate Large 7 Moraine dam condition (ice-cored/thermokarst, steeper slopes etc) None Minimum Partial moderate 8 Complex/compound threats None Slight Moderate. Large

Table 5: Hazard ranking (grading) on the basis of empirical scoring system

0 45 90 115 ≥140 Zero Minimal Moderate High Very high

GLOF can occur at any time

6.2 gloF Risk Assessment of Imja analyze the hazard potential in terms of GLOF glacier lake risk (Table 6)

The kind and combination of criteria used to It is not possible to make a complete assessment assess hazard potentials according to different of the GLOF risk without doing more detailed scale levels vary from case to case. Lake area investigation and research work of level 3. The and volume are of primary importance since above discussion and score analysis pertained to they define the amount of water available for an assess the risk of GLOF from Imja Lake based outburst. Analysis of Imja glacier lake based on on the current state-of-knowledge gained from the above empirical scoring system helped to remotely sensed data and previous reports.

Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal 88 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel December 2010

Table 6: Empirical scoring of GLOF hazard ) 2 Score threats Hazard Calving Volume channel Seepage area (m condition Rock fall/ Estimated Measured en-glacier Freeboard Compound Supra- and avalanches Lake Name volume (m3) Moraine dam

Imja glacier 959500 45609694 40 0 10 10 0 10 10 10 90 lake

It is not possible to make a complete assessment that the GLOF risk of the possible outburst from of the GLOF risk without doing more detailed Imja glacier lake is Moderate. It was found that investigation and research work of level 3. The there is a risk of GLOF from Imja Lake. The above discussion and score analysis pertained to condition may deteriorate faster in the future assess the risk of GLOF from Imja Lake based and GLOF risk might be higher. It is, therefore, on the current state-of-knowledge gained from suggested that regular routine investigation of key remotely sensed data and previous reports. area of imja glacier laske should be carried out. The results obtained from such investigation would 7. Conclusion help to update previously collected information and conditions of GLOF risk from Imja Glacier Determination of lake storage volume is one of can be analysed. Surrounding of Glacier lake and thematic area of glacier lake investigation. Direct its environment are found to be rapidly changing depth measurement method or echo-sounding are due to climate change, Imja glacier lake should used to determine lake storage volume. Based be monitored on yearly basis using very-high- on the limited available observed data of similar resolution satellite imagery (e.g. PRISM/IKONOS) nature and characteristics of glacier lake, volume and validation of remotely sensed lake area and storage has been estimated from observed data other physical surrounding condition should be of depth and area of glacier lakes by applying done through field investigation. statistical regression model. The comparison of observed and estimated lake volume showed that It is therefore rational to plan a mitigation there is a difference of ± 15 %.The functional strategy against Imja GLOF. The measure has relationship developed has been applied to Imja to be a combination of early warning system glacier lake to estimate lake volume storage from and lake lowering system. Detail scientific study the glacier lake area obtained by remote sensing on geo-technical, glaciological and hydrological images. aspects and continuous monitoring of Imja lake and its surroundings are essential to gather the It is concluded that the lake can be easily detected in-depth knowledge of the physical situation of and monitored by using remotely sensing data. the surrounding environment of the lake before Characteristics of the lake and the dam, and the deciding on the method to adopt. In view of potential for trigger mechanisms of lake outbursts the hazard potential of glacier lakes in Nepal can be evaluated using specific techniques on Himalayas, a systematic application of such remote different scale levels and derived criteria affecting sensing based methods is recommended. Based hazard/score. The result from this study showed on the observations of glacier lake surrounding

Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal December 2010 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel 89 including measured parameters of lake, a criteria Minimization of Potential GLOF Hazard based methodology can be established for from Imja Glacier Lake. carrying out the GLOF hazard potential grading which can be used for risk assessment of other Dwivedi, S.K., Acharya, M.D. and Simard, R. Glacier Lakes in similar environment. 2000. The Tam Pokhari Glacier Lake outburst floods of 3 September 1998. References Journal of Nepal Geological Society. 22. 539-546. Baban, S.M.J. 1993. The evaluation of different algorithms for bathymetric charting Fuji, Y. and Higuchi, K., 1976. Ground temperature of lakes using Landsat imagery. and its relation to permafrost occurences International Journal of Remote Sensing, in the Khumbu Region and Hidden Valley. 14: 2263–2274. Seppyo, 38: 125-128.

Benny, A.H., and Dawson, G.J. 1983. Satellite Fujita, K., Sakai, A., Nuimura, T., Yamaguchi, S. imagery as an aid to bathymetric charting and Sharma, R. R. 2009. Recent changes in the Red Sea. The Cartographic Journal, in Imja Glacier Lake and its damming 20: 5–16. moraine in the Nepal Himalaya revealed by in situ surveys and multi-temporal Braun, M. and Fiener, P. 1995 Report on the ASTER imager. IOP PUBLISHING, GLOF Hazard Mapping Project in the ENVIRONMENTAL RESEARCH Imja Khola/ Dudh Kosi Valley, Nepal, LETTERS (Environ. Res. Lett. 4 Contribution to the Project “Establishment (2009) 045205, 7pp), doi:10.1088/1748- of a Measuring Service for Snow and 9326/4/4/045205. Glacier Hydrology in Nepal”. Edited by: O.R. Bajracharya (Ag. Project Director) Grabs, W.E. and Hanisch, J. 1993. Objectives and W.E. Grabs (Team leader, GTZ). and prevention methods for glacier lake outburst floods (GLOFs). In Snow and Clague, J.J., and Evans, S.G. 1994. Formation and Glacier Hydrology. IAHS publication No. failure of natural dams in the Canadian 218. Edited by G.J. Young. 319-330. Cordillera. Geological Survey of Canada Bulletin 464, Ottawa. Costa, J.E., and Hanisch, J., Koirala, A., and Grabs, W.E. 1996. Schuster, R.L. 1988. The formation Ausbrüche von Gletscherseen in Nepal and failure of natural dams. Geological und ihre mögliche Verhinderung. Society of America Bulletin, 7: 1054– Zeitschrift für angewandte Geologie, 42: 1068. 8–13.

DHM, 1999. Monitoring the Growth Rate of Imja Huggle, C. Kääb, A. Haeberli, W. Paul, P. and Glacier Lake. Kathmandu Teysseire, P. 2002. Remote sensing based assessment of hazards from glacier lake DHM, 2001. Final Report on Preliminary outbursts: a case study in the Swiss Alps, Analysis for the Preparation Proposal for Can. Geotech. J. 39: 316-330.

Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal 90 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel December 2010 ICIMOD, 2001. Inventory of glaciers, glacier of the International Symposium on Snow lakes and glacier lake outburst floods, and Glacier Hydrology, Kathmandu.16-21 monitoring and early warning system in November 1992. Edited by G. J. Hindu Kush-Himalaya Region, Nepal, Young, International Association of ICIMOD/UNEP. Hydrological Sciences, Wallingford, England. IAHS Publication No. 218, pp. ICIMOD, UNEP, CAREERI, BHT, APN, START. 331–340. 2004. Inventory of Glaciers, Glacier Lakes, and the identification of potential Pietroniro, A., and Leconte, R. 2000. A review glacier lake outburst folds (GLOFs) of Canadian remote sensing applications affected by global warming in the in hydrology, 1995–1999. Hydrological mountains of Himalayan region, Poiqu Processes, 14: 1641–1666. (Bhote-Sun Koshi) and Rongxer (Tama Koshi) Basins of Tibet Autonomous RGSL, NEA. 2004. Consulting Services for Region, PR China. Feasibility Study of Upper Tama Koshi Hydroelectric Project. Appendix D: Ives, J.D. 1986. Glacier lake outburst floods Glacier Lake Outburst Flood (GLOF) and risk engineering in the Himalaya. Risk Assessment Study. Unpublished ICIMOD Occasional Paper 5, 42 pp. Technical Report J03170 Report v1.0, July 2004, Reynolds Geo-Sciences Ltd, Jakob, M. 1992. Blockgletscher in Khumbu Mold, UK. Himalaya, Nepal. Master Thesis, Geographischen Institut der Ruprecht- Richardson, S.D., and Reynolds, J.M. 2000. Karls-Universiät, 123pp. An overview of glacier hazards in the Himalayas. Quaternary International, Kääb A. 1996. Photogrammetrische Analyse 65/66: 31–47. zur Früherkennung gletscherund permafrostbedingter Naturgefahren im Sakai, A. 1995. Thermal Characteristics of a Hochgebirge. Mitteilungen der VAW/ glacier lake in Nepal Himalaya: Master’s ETHZ 145, Zürich. thesis submitted by, Graduate School of Environmental Earth Science, Hokkaido LIGG/WECS/NEA 1988. Report on First University, Japan. Expedition to Glaciers and Glacier Lakes in the Pumqu (Arun) and Poiqu (Bhote- Shrestha, A. B., 2007. Glacier Lake Outburst Sun Koshi) river basins, Xizang (Tibet), Flood Vulnerability Assessment. China, Sino-Nepalese Investigation of Himalayan Wetlands: Risk, Challenges Glacier Lake Outburst Floods in the and Opportunities, pp 27-29. Changwon: Himalaya. Beijing, China: Science Press Ramsar Wetlands Center Korea.

Meon, G., and Schwarz, W. 1993. Estimation of Shirahata, S., 1983. Nepal Himalaya. allied glacier lake outburst flood and its impact on Publishers Private Ltd., Bombay. 233pp. a hydro-project in Nepal. In Proceedings

Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal December 2010 K.P. Budhathoki, O.R. Bajracharya & B.K. Pokharel 91 Teysseire, P. 1999. Gefährliche Gletscherseen WECS, 1991. Preliminary Work Report on im Wallis. In Proceedings CENAT Glacier Lake Outburst Flood in the (Natural Hazards Competence Centre Nepal Himalaya, WECS Report No. (Switzerland)) Fachtagung Naturgefahren: 4/1/291191/1/1, Seq. No. 387, August Gletscher und Permafrost, Zurich, 29 1991, Kathmandu. March 1999, pp. 15–18. WECS . 1999. Water and Energy commission Vallon, M. 1989. Evolution, water balance, potential Secretariat Bulletin, vol 1, No 1 & 2, hazards and control of a pro-glacier lake April 1999, pp 11-13) in the French Alps. Annals of Glaciology, 13: 273–278. XU Daoming and FENG Qinghua. 1994. Dangerous glacier lakes and their Vuichard, D., and Zimmermann, M. 1987. The outburst features in the Tibetan 1985 catastrophic drainage of a moraine– Himalayas, Bulletin of Glacier Research dammed lake Khumbu Himal, Nepal: 12 (1994) 1-8, Data Center for Glacier cause and consequences. Mountain Research, Japanese Society of Snow Research and Development, 7: 91–110. and Ice.

Watanabe, T., Ives, J.D. and Hammond, J.E., Yamada T., 1998. Glacier Lakes and Their Outburst 1994. Rapid growth of a glacier lake in Floods in the Nepal. Data Center for Khumbu Himal, Himalaya: Prospects for Glacier Research, Japanese Society of a catastrophic flood. Mountain Research Snow and Ice. Monograph No. 1. March and Development, 14: 329-340. 1998. Tokyo.

WECS (1987b) Study of Glacier Lake Outburst Yongjian, D., and Jingshi, L. 1992. Glacier lake Floods in the Nepal Himalayas, Phase outburst flood disasters in China. Annals I,Interim Report, May, 1997, WECS of Glaciology, 16: 180–184 Report No. 4/1/200587/1/1, Seq. No. 251. Kathmandu, Nepal: WECS

Journal of Hydrology and Meteorology, Vol. 7, No. 1 SOHAM-Nepal